Performance Optimization of Closed-Field

Journal of Industrial Engineering and Sciences J. Indust. Eng. Sci.

KSIES

Volume (23), Issue (6), 2014, pp. 29-35

Performance Optimization of Closed-Field Unbalanced Dual Magnetrons DC Plasma Sputtering System Bahaa T. Chiad Department of Physics, College of Science, University of Baghdad, Baghdad, Iraq, [email protected]

Mohammad K. Khalaf Center of Applied Physics, Ministry of Science and Technology, Baghdad, Iraq, [email protected]

Firas J. Kadhim Department of Physics, College of Science, University of Baghdad, Baghdad, Iraq, [email protected]

Oday A. Hammadi Department of Physics, College of Education, Al-Iraqia University, Baghdad, Iraq, [email protected] (Received: 13 January 2014, Revised: 22 May 2014, Accepted: 24 June 2014, Published: 01 November 2014)

______________________________________________________________________________ In this work, the effects of home-made closed-field unbalanced dual magnetrons on the performance of plasma sputtering system were studied. These effects were introduced by comparing the obtained Paschen’s curve in existence and absence of magnetrons. The distribution of magnetic fields between the magnetrons was introduced to optimize the closed-field unbalanced configuration. Characterization of Paschen’s curve as well as discharge current with gas pressure at different distances between the discharge electrodes was introduced. Also, the current-voltage characteristics were introduced at the optimum operation conditions. © Korean Society for Industrial Engineering and Sciences Keywords: Physical vapor deposition, Sputtering, Magnetron sputtering, Paschen’s law

______________________________________________________________________________

1. Introduction A glow discharge is simply produced by applying an electric potential on a gas sample between two electrodes placed inside a vacuum chamber [1]. This glow discharge is a common source of plasma that can be established through an avalanche like ionization of gas neutrals at specific conditions for gas pressure and applied voltage [2]. At low kinetic energies (energies between 0 and about 50 eV), the ion does not have sufficient energy to dislodge the target atoms and thus the ejection of target particles occurs only for very special collision geometries [3-4]. With moderate energies (between 50 and roughly 1 keV), the ions impact dislodge “knock-on” atoms into the target, which by their turn will dislodge other target atoms [5]. Several studies showed that the ion energies must exceed four times the binding energy of the atoms of the target surface to induce sputtering. This induces collision cascade that eject atoms, ions, electrons and neutrals from the first 10 to 50Å of the surface of a target [6-9].

Initially existence of electrons is a requirement for this ionization process and such electrons, which may normally exist by cosmic radiation, are accelerated by the applied electric field towards the anode and consequently gained sufficient energy to collide with gas atoms and ionize them [8-9]. Hence, the number of electrons existing between the two electrodes will be multiplied at a rate described by the Townsend’s first ionization coefficient ( ), which represents the relative increase in electron flux (je) per unit path length as [10-12] (1) This equation can be solved as ( ) ( ) (2) where je(0) is the current density from electron current emitted from the cathode at x=0 The continuous electron production is required to sustain the discharge and provided by the production of secondary electrons induced during the ion impact at the cathode. The electron production can be

© 2014 All rights reserved for Korean Society for Industrial Engineering and Sciences (KSIES) Available online at http://jieseditor.wix.com/jieswebsite#!contact/cvrh

29

Characteristics and Operation Conditions of a Closed-Field Unbalanced DC Magnetron Plasma Sputtering System

described by the emission coefficient (e) of the secondary electrons as: ( ) ( ) (3) For the case of producing one ion in the cascade with each produced electron, we have ( ) ( ) ( ) ( ) ( ) ( ) ( )( ) (4) Here,  is equivalent to the product of the net ionization probability per unit length and the probability of ion generation by the collision and can be written as [13] (

)

(5)

where  is the collision mean free path and eE is the energy gained by electrons between consecutive collisions Knowing that the mean free path is inversely proportional to the gas pressure, the Townsend’s coefficient can be given by ( ) (6) where A and B are constants and their values are determined by the properties of the used gas, p is the gas pressure, d and V are the distance and the applied voltage between discharge electrodes as the electric field (E) is included by the ratio V/d. When the electron multiplication gets very efficient and je(d)>>je(0), a breakdown condition is reached. If the term of the ion flux at the anode in Eq. (4) is neglected, then the following condition is obtained (

)

(7)

By treating both equations (6) and (7) mathematically, the breakdown voltage is given by (8) (

)

(

(

))

and this equation is known as Paschen’s law. The breakdown voltage (VB) depends on the product of pressure (p) and electrode separation (d) as this product is denoted as “pd”, while this voltage weakly depends on the cathode material that defines the emission coefficient of secondary electrons. As well, the breakdown voltage is proportional to the product pd at large values of this product and the electric field (E=V/d) is scaled linearly with the pressure [14]. In case of small values of the product pd, only few collisions occur and higher voltage is applied to increase the probability of breakdown per collision. Hence, the minimum voltage required to ignite the discharge of a gas sample of pressure p over a distance d is defined at the minimum of Paschen’s curve, where | ( ) (9) If the pressure and/or separation distance is too large, ions generated in the gas are slowed by inelastic collisions so that they strike the cathode with insufficient energy to produce secondary

30

electrons. In most sputtering glow discharges, the discharge starting voltage is relatively high. Eventually, an avalanche occurs in which the ions striking the cathode release secondary electrons, which form more ions by collision with neutral gas atoms. These ions then return to the cathode, produce more electrons. When the number of electrons generated is just sufficient to produce enough electrons to regenerate the same number of electrons, the discharge is self-sustaining. The gas begins to glow, the voltage drops, and the current rises abruptly [15]. This is called the “normal glow”. The color of this luminous region is characteristic of the excitation gas used. Since the secondary electron emission ratio of most materials is of the order of 0.1, more than one ion must strike a given area of the cathode to produce another secondary electron. The bombardment of the cathode in the normal glow region self-adjusts in the area to accomplish this. Initially, the bombardment is not uniform, but concentrated near the edges of the cathode or at other irregularities on the surface. As more power is supplied, the bombardment increasingly covers the cathode surface until a nearly-uniform current density is achieved [14]. A further increase in power produces both increased voltage and current density in the discharge. This abnormal glow is the mode used for sputtering [16]. If the cathode is not cooled, at a current density of approximately 0.1A/cm2, thermionic electrons are emitted in addition to secondary electrons, followed by a further avalanche. The output impedance of the power supply limits the voltage and the low-voltage highcurrent arc discharge forms. The abnormal glow is dependent on the breakdown voltage VB. This in turn is dependent on the mean-free-path of secondary electrons and the distance between the anode and the cathode. Each secondary electron must produce about 10-20 ions for the original avalanche to occur. If gas pressure is too low, or the cathode-anode separation too small, the secondary electrons cannot undergo a sufficient number of ionizing collisions before they strike the anode [17]. Secondary electrons are repelled at high velocity from the cathode and start to make collisions with neutral gas atoms at a distance away from the cathode corresponding to their mean free path. This leaves a dark space which is very well defined [18]. Since the electrons rapidly lose their energy by collisions, nearly all of the applied voltage appears across this dark space. The dark space is also the region in which positive ions are accelerated toward the cathode [19]. Since the mobility of ions is very much less than that of electrons, the predominant species in the dark space are ions. Acceleration of secondary electrons from the cathode results in ionizing collisions in the negative glow region [20].

J. Indust. Eng. Sci. Volume (23), Issue (6), 2014, pp. 29-36

2. Experiment The main parts of the plasma sputtering system are shown in Fig. (1) and the closed-field unbalanced magnetron (CFUBM) was employed at the cathode electrode. Figure (2) shows the design of closed-field unbalanced magnetron (CFUBM) used in this work. Electrodes (anode and cathode) were made of stainless steel and each was a disk of 8 cm in diameter and 4 mm in thickness. Two annular concentric magnets were placed behind each electrode to form the dual magnetron configuration. The outer diameters of the two magnets were 8 cm and 4 cm, while the inner diameters were 4 cm and 3.2, respectively.

Fig. (1) Schematic diagram of the system used in this work (above) and a photograph of the magnetron employed at each electrode (below)

Fig. (2) Design of closed-field unbalanced magnetron (CFUBM) used in this work

The electrodes were connected to a DC power supply to provide the electrical power required for discharge. The lower electrode (anode) could be move vertically with respect to the fixed upper electrode (cathode) to adjust the separation of the two electrodes from 1 to 8 cm. Pure argon gas was used to produce the discharge plasma for sputtering. A DC power supply providing voltage up to 6kV was used for electrical discharge between the electrodes and both breakdown voltage (up to 1 kV) and discharge current (up to 100 mA) were monitored by two digital voltmeter and ammeter, respectively. The discharge chamber was evacuated by a two-stage Leybold-Heraeus rotary pump and the vacuum inside chamber was measured by Pirani gauge connected to a vacuum controller from Balzers VWS 120. Argon gas was supplied to the chamber through a fine-controlled needle valve (0-160 ccm) to control the gas pressure inside the chamber. The magnetic field intensity of the magnetron was measured by a Hall probe connected to a teslameter (model NV621). First, the magnetic field intensity of each magnetron was measured individually, and second, the magnetic field intensity between both magnetrons was measured too. In the second case, the magnetrons were maintained parallel to each other and the probe was positioned between them using an adjustable clamp. Measurements were carried out over all the distance between the two magnetrons in all coordinates in order to determine the spatial distribution of the magnetic field. 3. Results and Discussion In order to determine the optimum distance between two magnetrons, the magnetic field intensity was measured at the midpoint along the distance between the two magnetrons and at 2.2 cm from the edge of the electrode and the results are shown in Fig. (3). The maximum was observed at 4 cm, which can be considered as the optimum distance, while the minimum was measured at ≥ 8 cm. The diameter of the measuring probe is 0.8 cm, therefore, the minimum distance was 2 cm in order to locate the measuring probe at the midpoint. The behavior shown in the figure below is attributed to the interference between the lines of the magnetic field, i.e., the maximum interference occurs at the midpoint of 4cm separation, whereas this interference decreases as the two magnetrons move away from each other reaching to “no interference” condition at ≥ 8 cm separation. The electrons are unable to travel perpendicular to the magnetic field lines over distances greater than Larmor radius, therefore, they are confined. The electric field on the other hand causes the

31

Characteristics and Operation Conditions of a Closed-Field Unbalanced DC Magnetron Plasma Sputtering System

electrons to move in the direction perpendicular to both the electric field and the magnetic field (E×B or Hall drift). The combination of the electron confinement and the E×B drift ensures that the electrons have a much longer mean free path in the plasma than in conventional glow discharges, giving rise to more ionization collisions, and consequently higher ion fluxes. These ion fluxes are highest in between the magnets, hence most of the target material is sputtered there. This gives a characteristic feature of conventional planar magnetrons called the racetrack. This racetrack generally limits the complete target utilization, resulting in higher working costs. This problem can be overcome by using rotatable magnetrons. Instead of a cylindrical inner magnet and an outer magnet ring, these magnetrons consist of a central bar shaped magnet surrounded by a rectangular shaped magnet configuration around which a cylindrical target rotates. This greatly enhances the utilization of the target and is therefore much more interesting for industrial applications.

decreased to seventh its value in absence of magnetron. However, Paschen’s curves of both cases are identical with different minima, as the value of “p.d” product was 1.1 mbar.cm when no magnetron is used and 1.4 mbar.cm when a magnetron is used. In dc sputtering, the electrons that are ejected from the cathode are accelerated away from the cathode and are not efficiently used for sustaining the discharge [4,19]. To avoid this effect, a magnetic field is added to the dc sputtering system that can deflect the electrons to near the target surface, and with appropriate arrangement of the magnets, the electrons can be made to circulate on a closed path on the target surface. This high current of electrons creates high-density plasma, from which ions can be extracted to sputter the target material, producing a magnetron sputter configuration [20].

Fig. (4) Paschen’s curve for the constructed plasma sputtering system with and without magnetron

Fig. (3) Variation of magnetic field intensity along the vertical distance separating the two magnetrons The sputtering electrons are assumed to be trapped near the cathode by the magnetic field of the magnetron in order to increase the path length of these electrons. Therefore, the maximum interference between the two fields is not preferred for such purpose because the electrons would not be seized near the cathode. Instead, these electrons may be drifted by the interfered lines away from the cathode. Sputtering of a target atom is just one of the possible effects resulting from the surface ion bombardment. Aside from sputtering, the second important process is the emission of secondary electrons from the target surface, which play a fundamental role in keeping the sputtering process itself. Figure (4) shows Paschen’s curve for both cases of using and not using the magnetron at the upper electrode (cathode). As clearly shown, the effect of using magnetron lies in decreasing the breakdown voltage to about 15% of its initial (maximum) value, while the minimum voltage was

32

A disadvantage of the magnetron sputtering configuration is that the plasma is confined near the cathode and is not available to active reactive gases in the plasma near the substrate for reactive sputter deposition. This difficulty can be overcome using an unbalanced magnetron configuration, where the magnetic field is such that some electrons can escape from the cathode region [21]. An unbalanced magnetron (UBM) has a proper magnetic field configuration in which a finite degree of the field lines from the outer magnetic pole diverge to the substrate, though the rest of the lines finish on the inner pole behind the target. Sufficient plasma density and a positive ion current on a metallic substrate even at a large distance from the target can be achieved in the unbalanced magnetron as compared with the balanced one [22]. The plasma sputtering system was then characterized by the relation of discharge current to the gas pressure inside the chamber at different inter-electrode distances, as shown in Fig. (5). Again, all curves are identical with the discharge current shifted upward on the vertical axis. According to Eq. (2), as the distance between the electrodes is decreased, the current density from

J. Indust. Eng. Sci. Volume (23), Issue (6), 2014, pp. 29-36

Fig. (5) Variation of discharge current with increasing gas pressure for different inter-electrode distances (d) By modeling the data obtained from this work, we plotted Paschen’s curves for different separation distances between the discharge electrodes, as shown in Fig. (6), in order to introduce how far the experimental data deviate from the optimum behavior determined by the analytical equations.

Fig. (6) Modeling of Paschen’s curves obtained at different inter-electrode spacing distances Experimental data are lower than expected by analytical equations due to the experimental restrictions, such as measuring accuracy and thermal effects inside discharge chamber and plasma region. As well, the modeling process is based on sharply consecutive output data obtained from analytical equations, i.e., no deviation may be expected in the behavior of the modeling formulae. This may submit an advantage to those designing, operating or working on such systems to take into account the optimum behaviors extracted from similar experimental studies. In order to introduce of using magnetron, the same relation between discharge current and discharge voltage was recorded at certain interelectrode distance of 4cm as shown in Fig. (7). Despite the discharge current measured at this distance was lower than that measured at 5cm, this distance was chosen for this relation to match the optimum distance of magnetron configuration, as shown before. It is clear that using dual magnetrons results in higher discharge current at certain discharge voltage when compared to the case of using only one magnetron (at the cathode), which in turn shows higher current when compared to no magnetron case. This effect is attributed to the role of magnetron in confining charged particles (electrons and ions) near the electrodes, increasing number of collisions with neutral gas atoms and hence producing much more particles those carry much more charges in time (higher current). Using only one magnetron at the cathode results in an increase of about 13% in discharge current when compared to the case of no magnetron used, while using dual magnetrons at both electrodes results in an increase of about 24% when compared to the case of no magnetron used. 100

Discharge current (mA)

electron current emitted from the cathode je(0) is increased because less number of electrons are able to reach the anode and hence lower current flows. However, compensation is required between gas pressure and distance to work at a given discharge current before converting into decreasing current as saturation is reached. The plasma pressure will also have an impact on the sputtering process. Higher pressure means high density of gas molecules in the chamber. As a consequence, this will lead to higher electron density and ion density in the plasma. The high electron and ion density will increase the bombardment counts and also the probability of collision between particles. The applied voltage will have an impact on the charged particles density and the energetic level of the charged particles. The density of ions is mainly related with the power input to the plasma. Usually, one thirtieth of the discharge energy will be transformed into ionization process. However, the applied voltage shouldn’t reach too high for avoiding the implantation of ions.

No magnetron One magnetron Dual magnetrons

80

60

40

20

d=4cm 0 200

250

300 350 400 450 Discharge voltage (V)

500

550

Fig. (7) Current-voltage characteristics of discharge at inter-electrode distance of 4cm when no

33

Characteristics and Operation Conditions of a Closed-Field Unbalanced DC Magnetron Plasma Sputtering System

magnetron used, only one magnetron used and dual magnetrons used In order to introduce the effects of using magnetrons on the parameters of plasma, the Langmuir probe measurements were performed when only one magnetron used at the cathode and when dual magnetrons used and then compared the obtained results to those obtained when no magnetron used, as shown in Fig. (8). It is clear that using only one magnetron at the cathode caused to decrease the drained current by the probe by about 12%. This decrease may be attributed to the role of this magnetron in trapping a fraction of the electrons in discharge volume near the cathode to increase the ionization rate of the neutral gas atoms and hence preventing the probe from collecting more electrons. Probe current (mA)

100 80 60 40

No magnetron One magnetron Dual magnetron

20 0

-100

-50

0 -20

50

100

Bias voltage (V)

Fig. (8) Current-voltage characteristics of Langmuir probe diagnostics in glow-discharge plasma at three different cases (no magnetron, using one magnetron at the cathode, and using dual magnetrons) When dual magnetrons were used, the drained current by the probe was decreased by about 16% due to the roles of both magnetrons in trapping much more charged particles near the cathode and anode and hence reducing the number of particles passing the distance between the electrodes where the probe is placed. However, these roles could not prevent the discharge current from flowing between the electrodes but even these particles sustaining the discharge are accelerated by the both electric and magnetic fields to higher drift velocities that the probe could not attract them from their paths across the inter-electrode distance [23]. Accordingly, the magnetrons caused to change the values of plasma parameters, such as electron and ion temperatures and densities, as these parameters are deduced from the current-voltage characteristics of the probe immersed in plasma. These variations are observed in Table (1). Table (1) Effect of using magnetrons on the plasma parameters

34

Electron temperature (eV) Electron density (x1021m-3) Ion temperature (eV) Ion density (x1021m-3)

No magnetron

One magnetron

Dual magnetrons

4.856

4.871

4.850

1.329

1.148

1.130

0.889-1.192

0.888-1.189

0.864-1.159

1.329

1.148

1.130

These measurements were repeated when the probe was placed near each of the electrodes (cathode and anode) when dual magnetrons were used in order to introduce the plasma parameters there, as shown in the Fig. (9). The probe current was higher than other positions because the densities of electrons near the cathode are higher than any other region inside plasma since the cathode is the source of discharge electrons in addition to the trapping effect of the magnetron placed at the cathode. Therefore, the probe would collect much more electrons in this region. The increase in probe current near the cathode was about 46% than its value at the center point between the electrodes. The probe current near the anode was higher by about 6% than its value at the center point because the density of electrons near the anode is already low and the magnetron at the anode also plays a role in trapping a fraction of these electrons near the anode. Therefore, the number of electrons collected by the probe near the anode is reasonably lower than that that near the cathode [34]. 120 Probe current 100 (mA) 80 60 Center Near Cathode Near Anode

40 20 0

-100

-50

0

50

100

-20 Bias voltage (V)

Fig. (9) Current-voltage characteristics of Langmuir probe diagnostics in glow-discharge plasma at three different positions of the Langmuir probe inside plasma volume Finally, the system considered in this work was used for deposition of metal oxide thin films (e.g., stainless steel films on glass substrates) and these films were featured by uniformity, very good adhesion and large area when compared to other deposition techniques available in the local research environment, therefore, these films could be

J. Indust. Eng. Sci. Volume (23), Issue (6), 2014, pp. 29-36

described as high quality structures. However, the results of deposition were not included in this work as they are presented in other studies carried out using the same system. 4. Conclusion Referring to the results obtained from this work, the home-made dc magnetron plasma sputtering system was characterized to introduce its performance in accordance to magnetic field distribution, electrical characteristics, Paschen’s law and governing properties of such deposition systems. Results have showed that using magnetron at the cathode of discharge configuration highly has affected these operation characteristics as the breakdown voltage was decreased, the minimum point was little shifted upward, . This system was found to satisfy the requirements for deposition of high-quality thin films from different materials.

[16] T.E. Sheridan, M.J. Goeckner, and J. Goree, J. Vac. Sci. Technol. A 16(4) (1998) 2173-2176. [17] F. Papa et al., Thin Solid Films, 520(5) (2011) 15591563. [18] B. Liebig et al., Surf. Coat. Technol., 205 (2011) S312–S316. [19] P.J. Kelly and R.D. Arnell, Surf. Coatings Technol., 112 (1999) 170-176. [20] A.S. Penfold, in Handbook of Thin Film Process Technology, IOP (1995), sec. A.3.2, p. 1. [21] B. Window and N. Savvides, J. Vac. Sci. Technol., A 4(2) (1986) 196. [22] T. Makabe and Z. Petrovic, Plasma Electronics: Applications in Microelectronic Device Fabrication, Taylor & Francis, New York (2006) 301. [23] B.T. Chiad, M.K. Khalaf, F.J. Kadhim, O.A. Hammadi, Characteristics and Operation Conditions of a Closed-Field Unbalanced DC Magnetron Plasma Sputtering System, J. Ind. Eng. Sci., accepted for publication, to appear in 2015.

_________________________________________ Acknowledgement Authors would like to appreciate the valuable assistance submitted by Dr. Fuad T. Ibrahim to perform the machining work of the proposed design. References [1] E.D. McClanahan and N. Laegreid, Topics in Appl. Phys., 64 (1991) 339-377. [2] D. Rademacher, T. Zickenrott, M. Vergöhl, Thin Solid Films 532 (2013) 98–105. [3] J.-S. Baek and Y.J. Kim, Cooling Effect Enhancement In Magnetron Sputtering System, 5th Inter. Conf. on CFD in the Process Industries CSIRO, Melbourne, Australia 13-15 December 2006, 1-5. [4] P.J. Kelly and R.D. Arnell, Vacuum 56 (2000) 159172. [5] R.D. Arnell and P.J. Kelly, Surf. Coat. Technol., 112 (1999) 170–176. [6] J. Walkowicz et al., Tribologia, 6 (2006) 163-174. [7] D.E. Ashenford et al., Surf. Coat. Technol., 116-119 (1999) 699-704. [8] P. Sigmund, Topics in Applied Physics: Sputtering by Particle Bombardment I, ed. R. Behrisch, Vol. 47, Springer-Verlag (Berlin) 1981. [9] F. Taherkhani and A. Taherkhani, Transaction B: Mechanical Engineering, 17(4) (2010) 253-263. [10] G.K. Wehner and G.S. Anderson, Vacuum Technology, Thin Films and Sputtering: An Introduction, ed. R. V. Stuart, Academic Press, New York, 1983. [11] A. Bogaerts et al., Spectrochimica Acta Part B 57 (2002) 609–658. [12] E.F. Kotp and A.A. Al-Ojeery, Australian J. Basic Appl. Sci., 6(3) (2012) 817-825. [13] S.D. Personick, Bell System Tech. J., 50(10) (1971) 3075-3095. [14] J.-C. Wang et al., J. Appl. Phys., 113 (2013) 033301 [15] L.M. Isola, B.J. Gomez and V. Guerra, J. Phys. D: Appl. Phys., 43 (2010) 015202.

Bahaa T. Chiad He was born in Babylon, 1954, received his B.Sc. in Physics, M.Sc. in Astrophysics and PhD in Molecular Physics from the University of Baghdad, Baghdad (Iraq) in 1983, 1988 and 1993, respectively. He is currently a professor of molecular physics at the Department of Physics, College of Science, University of Baghdad. Mohammad K. Khalaf He was born in Baghdad, 1970, received his B.Sc., M.Sc. degrees in Physics and PhD degree in Molecular Physics from the University of Baghdad, Baghdad (Iraq) in 1990, 1998 and 2010, respectively. He is currently senior researcher at the Applied Physics Center, Ministry of Science and Technology, Baghdad (Iraq). Firas J. Kadhim He was born in Baghdad, 1971, received his B.Sc., M.Sc. degrees in Physics and PhD degrees in Molecular Physics from the University of Baghdad, Baghdad (Iraq) in 1993, 2000 and 2009, respectively. He is currently assistant professor at the Department of Physics, College of Science, University of Baghdad. Oday A. Hammadi He was born in Baghdad (Iraq), 1975, received his B.Sc. and M.Sc. degrees in Laser Physics from the University of Technology, Baghdad (Iraq) in 1997 and 2000, respectively, and PhD degree in Molecular Physics from the University of Baghdad, Baghdad (Iraq) in 2014. He is currently a lecturer at the Department of Physics, College of Education, Al-Iraqia University, Baghdad (Iraq).

__________________________________________________________________________________________

35